INTRODUCTION

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

Calcium is a ubiquitous second messenger in eukaryotic signal transduction cascades. In plants, intracellular Ca²⁺ levels are modulated in response to various signals, including hormones, light, mechanical disturbances, abiotic stress, and pathogen elicitors (Sanders et al., 1999; Evans et al., 2001; Rudd and Franklin-Tong, 2001). How response specificity is regulated during Ca²⁺-mediated signal transduction is an important biological issue. It appears that different stimuli elicit specific calcium signatures, generated by altering the kinetics, magnitude, and cellular source of the influx (Malhó et al., 1998; Allen et al., 2000, 2001; Evans et al., 2001; Rudd and Franklin-Tong, 2001). Unlike most other ions, calcium does not freely diffuse within cells (Trewavas, 1999). Plants have multiple calcium stores, including the apoplast, vacuole, nuclear envelope, endoplasmic reticulum (ER), chloroplasts, and mitochondria. Therefore, each stimulus can elicit a characteristic Ca²⁺ wave by specifically altering the activities of various differentially localized Ca²⁺ channels, H⁺/Ca²⁺ antiporters, and Ca²⁺- and H⁺-ATPases (Bush, 1995; Thuleau et al., 1998; Allen et al., 2000; Hwang et al., 2000; Harper, 2001). Different calcium sensors recognize specific calcium signatures and transduce them into downstream effects, including altered protein phosphorylation and gene expression patterns (Sanders et al., 1999; Rudd and Franklin-Tong, 2001).

In plants, there are several known classes of Ca²⁺-binding sensory proteins, including calmodulins, calcium-dependent protein kinases (CDPKs), and calcineurin B-like proteins. Among them, CDPKs are the best characterized and are of particular interest. They represent a novel class of Ca²⁺ sensors, having both protein kinase and calmodulin-like domains in a single polypeptide. As a result, CDPKs directly bind calcium, and their calcium-stimulated kinase activities are independent of calmodulins, unlike calcium/calmodulin-dependent protein kinases (CaMKs; Roberts and Harmon, 1992). Currently, most of the known calcium-stimulated protein kinase activities in plants are associated with CDPKs. A genome-wide analysis of Arabidopsis CDPKs provides an overview of the diversity of this large multigene family and should facilitate the elucidation of their functions. It appears likely that gene duplication and subsequent evolution generated CDPKs with both redundant and distinct functions. Furthermore, the functional specificity of individual CDPKs may be determined by regulations at both transcriptional and posttranslational levels, as well as targeted subcellular compartmentalization, calcium and lipid sensitivity, and substrate recognition.

	THE ARABIDOPSIS CDPK GENE FAMILY

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

The first calcium-dependent, calmodulin-independent protein kinase activities were reported in pea (Pisum sativum) extracts 20 years ago (Hetherington and Trewavas, 1982). Since their initial purification and characterization from soybean (Glycine max; Harmon et al., 1987), CDPKs have been identified throughout the plant kingdom from green algae to angiosperms (Hrabak, 2000; Harmon et al., 2001). Other than plants, CDPKs are found only in some protozoans, and are notably absent from the sequenced eukaryotic genomes of yeast (Saccharomyces cerevisiae), nematodes (Harmon et al., 2000), fruitflies (Drosophila melanogaster; Adams et al., 2000), and humans (Homo sapiens; International Human Genome Sequencing Consortium, 2001; Venter et al., 2001). Analysis of the genome sequence of Arabidopsis indicates the presence of 34 CDPK genes (The Arabidopsis Genome Initiative, 2000). Information available from limited genomic sequencing, as well as several extensive expressed sequence tag (EST) projects, also indicate the presence of multigene families of CDPKs in other plants, including soybean, tomato (Lycopersicon esculentum), rice (Oryza sativa), and maize (Zea mays; Harmon et al., 2001).

Domain Structure

Four distinct domains typify CDPK family members: an N-terminal variable domain, a protein kinase domain, an autoinhibitory domain, and a calmodulin-like domain (Fig. 1). Based on phylogenetic analysis, it is thought that the CDPK gene family arose through the fusion of a CaMK and a calmodulin (Harper et al., 1991; Suen and Choi, 1991; Harmon et al., 2000; Zhang and Choi, 2001). This unique molecular structure allows the direct activation of CDPKs by Ca²⁺. Unlike the analogous mammalian protein, the multisubunit CaMKII, CDPKs function as monomers (Roberts and Harmon, 1992).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Structural comparisons of mammalian CaMKII and plant CDPKs. The kinase domain of CDPKs is up to 44% identical (65% similar) to that of mouse (Mus musculus) CaMKII alpha (accession no. S04365) and 43% identical (65% similar) to that of mouse CaMKII beta (accession no. NP_031621). N, Amino-terminal variable domain; K, kinase domain; A, autoinhibitory domain; CaM, calmodulin. The four bars within the CaM-like domain represent the EF hand Ca2+-binding sites.

Alignments of the predicted amino acid sequences of all 34 Arabidopsis CDPKs reveal a high conservation of the kinase (44%-95% identity and 60%-98% similarity), autoinhibitory (23%-100% identity and 42%-100% similarity) and calmodulin-like (27%-97% identity and 50%-98% similarity) domains, whereas the N-terminal variable domain shows little sequence similarity. (The alignments and pair-wise comparisons of the amino acid sequences of all 34 Arabidopsis CDPKs can be viewed online at http://xanadu.mgh.harvard.edu/sheenweb/index.htm by selecting "Arabidopsis CDPKs.") The kinase domain (264-273 amino acids long) contains all 12 of the highly conserved subdomains of typical eukaryotic Ser/Thr protein kinases (Hanks and Hunter, 1995). Nearly 100% identity is found in the region of the active site among all 34 Arabidopsis CDPKs. Using casein, histone IIIS, or syntide as substrates, heterologously expressed CDPKs have been shown to possess Ca²⁺-stimulated protein kinase activity in vitro (e.g. Harper et al., 1993; Urao et al., 1994; Lee et al., 1998; Yoon et al., 1999). An absolutely conserved Lys residue located within subdomain II is thought to be the ATP-binding site, and mutagenesis of this residue abolishes the catalytic activities (Sheen, 1996).

The autoinhibitory domain is a basic amino acid region (31 amino acids long) that functions as a pseudosubstrate (Harmon et al., 1994). Sixteen of the 34 Arabidopsis CDPKs contain a potential autophosphorylation site (Basic-X-X-S/T, where X is any residue) in the autoinhibitory domain (Harmon et al., 1994). Whether these CDPKs autophosphorylate at this site has yet to be determined. Although autophosphorylation at the analogous site of CaMKII results in a constitutively active enzyme that is no longer dependent upon Ca²⁺/calmodulin for activation (Schulman and Lou, 1989), it has not been established that an analogous phosphorylation plays a role in the activity of plant CDPKs (see "Regulation by Phosphorylation and Dephosphorylation" below).

The calmodulin-like domain (94-147 amino acids long) contains Ca²⁺-binding EF hands allowing the protein to function as a Ca²⁺ sensor. Each EF hand consists of a loop of 13 amino acid residues flanked by two -helices. A single Ca²⁺ molecule is bound to each EF hand via the loop domain (Zhang and Yuan, 1998). The number of EF hands differs depending on the isoform. Most Arabidopsis CDPKs contain four EF hands, whereas a few of them have one, two, or three (Table I). The most conserved EF hand sequences are those of the hands in positions 1 and 2 and the least conserved is that for position 4. The positions where the EF hands are absent also vary. These differences in numbers and positions of EF hands likely yield variations in the allosteric properties of Ca²⁺ binding and the activation threshold. Sequential deletion of the EF hands demonstrates that the number of EF hands may be important for determining calcium regulation of CDPK activity (Hong et al., 1996). In addition, site-directed mutagenesis of a highly conserved Glu residue in each EF hand shows that the closer the EF hand is to the autoinhibitory domain, the greater its effect on the Ca²⁺ regulation of CDPK activity (Zhao et al., 1994).

The mechanism by which CDPK activity is regulated is largely controlled through interactions between the kinase, autoinhibitory, and calmodulin-like domains. Under the basal condition of low free Ca²⁺, the autoinhibitory domain is bound by the kinase domain, keeping substrate phosphorylation activity low. Upon binding Ca²⁺ via the EF hand motifs, CDPKs undergo conformational changes that release the pseudosubstrate from the catalytic site, activating the protein (Harmon et al., 1994; Harper et al., 1994). This model is supported by the observation that deletion of only the calmodulin-like domain generates an inactive enzyme that cannot be activated by Ca²⁺, whereas deletion of both the autoinhibitory and calmodulin-like domains creates a constitutively active, Ca²⁺-insensitive enzyme (Harper et al., 1994; Sheen, 1996).

Little is known about the function of the N-terminal variable domain. It has been proposed that this region contains subcellular targeting information (Schaller and Sussman, 1988; Harper et al., 1994; Hrabak et al., 1996). CDPKs are reported to associate with various membranes (Ellard-Ivey et al., 1999; Martin and Busconi, 2000; Lu et al., 2001; Lu and Hrabak, 2002). However, none of the 34 Arabidopsis CDPKs are predicted to be integral membrane proteins. The N-terminal leader sequence of CDPKs is variable not only in amino acid sequence, but also in length, ranging from 25 (AtCPK11) to nearly 200 (AtCPK2) amino acids in Arabidopsis (Table I). Despite this variability within the N-terminal domain, most Arabidopsis CDPKs have a Gly residue at the second position. When placed in a proper context, this N-terminal Gly residue can be modified by covalent attachment of myristic acid (a C 14:0 fatty acid; Towler et al., 1988). In many systems, N-myristoylation promotes protein-membrane and protein-protein interactions (Johnson et al., 1994). Mutation of the N-terminal Gly abolishes such a lipid modification and thus prevents membrane association (Martin and Busconi, 2000). Twenty-four of the Arabidopsis CDPKs are predicted to have N-myristoylation motifs for membrane association (Table I). Among them, however, only AtCPK2 has been shown experimentally to be myristoylated at the N-terminal Gly residue, and the first 10 amino acids are critical for localization to the ER membrane (Lu and Hrabak, 2002; Table I). This lipid modification also has been shown to occur in CDPKs from four other species, and similarly to Arabidopsis, two of these proteins have been found to be membrane associated (Farmer and Choi, 1995; Ellard-Ivey et al., 1999; Martin and Busconi, 2000; Raíces et al., 2001). AtCPK 5 and 6 are not predicted to be myristoylated (Table I), but are partially associated with membranes nevertheless (Lu et al., 2001), suggesting that other mechanisms (e.g. glycosylation) may account for their subcellular localization.

The addition of a myristic acid residue is not always sufficient for membrane attachment. Often, a second lipid modification, such as palmitoylation (addition of palmitate, a C 16:0 fatty acid), is necessary to stabilize the interaction with the membrane. All 24 AtCPKs predicted to have a myristoylation consensus sequence also have at least one Cys residue at position 3, 4, or 5 (Table I), a potential palmitoylation site (Milligan et al., 1995). Recently, both myristoylation at the N-terminal Gly and palmitoylation at the Cys residues in positions 4 and 5 have been shown experimentally to occur in the membrane-bound rice OsCPK2 (Martin and Busconi, 2000). When myristoylation of OsCPK2 is abolished by mutating the N-terminal Gly, the protein can no longer be palmitoylated either, indicating that myristoylation may be a prerequisite for palmitoylation. Whether Arabidopsis CDPKs also have such a prerequisite, however, remains to be determined.

Sequence Homology and Chromosomal Distribution

All 34 Arabidopsis CDPKs are highly homologous to each other. Pair-wise analyses with the full protein sequences indicate that the overall identities and similarities are 39% to 95% and 56% to 96%, respectively. High identities are found between AtCPK 4 and 11 (95%), AtCPK 17 and 34 (93%), AtCPK 7 and 8 (90%), AtCPK 10 and 30 (86%), AtCPK 9 and 33 (85%), AtCPK 1 and 2 (81%), and AtCPK 21 and 23 (81%), and among AtCPK 5, 6, and 26 (85%-88%). Because both AtCPK 10 and 30 specifically activate a stress pathway (Sheen, 1996), such high homologies may indicate similar functions. AtCPK16, 18, and 28 are the most divergent CDPKs, as indicated by their relatively low average pair-wise identity/similarity values (45% and 64%, respectively).

To examine protein relationships of Arabidopsis CDPKs further, an unrooted tree was constructed from alignments of the full CDPK sequences (Fig. 2). Based upon sequence homology, the CDPKs of Arabidopsis cluster into four subgroups (I-IV). Subgroup IV is the least complex, with three members, and subgroup II is the most complex, with 13 members. This pattern of grouping was also found when the tree was constructed based on the sequences of the kinase domain only (data not shown; Harmon et al., 2001). Subgroups I through III are closer in sequence identity to each other than to subgroup IV. It is not known whether such a pattern of clustering reflects any functional differences between the subgroups.

View larger version (123K): [in this window] [in a new window]   Figure 2.   Relatedness of Arabidopsis CDPKs. The complete protein sequences of the AtCPKs were aligned and analyzed by the Treeview 1.6.5 program (http://taxonomy.zoology.gla.ac.uk/rod/rod.html). The unrooted distance tree reveals the presence of four distinct, branched subgroups (I-IV). The branch lengths are proportional to divergence, with the scale of "0.1" representing 10% change.

The 34 Arabidopsis CDPKs are distributed among all five chromosomes (Fig. 3). Chromosome IV has the most CDPKs (11), whereas chromosome III has the least (4). The only region that contains no CDPKs is the short arm of chromosome II. Interestingly, one gene cluster on the short arm of chromosome IV contains five genes (AtCPK 21, 22, 23, 27, and 31), all within subgroup IV. They are organized in tandem in the same transcriptional orientation (Fig. 3), and their amino acid sequences are very homologous (61%-82% identity and 74%-89% similarity). Furthermore, sequence homology also exists in the N-terminal variable domain in this gene cluster (21%-78% identity and 22%-85% similarity). These results suggest that they arose relatively recently by gene duplication and that they may have similar or overlapping functions.

View larger version (28K): [in this window] [in a new window]   Figure 3.   Genomic distribution of CDPKs on the Arabidopsis chromosomes. Ovals on the chromosomes indicate the location of the centromeres. The arrows next to the gene names show the direction of transcription. The numbers in parentheses designate the position of the first exon of each CPK gene in megabases (Mb). The chromosome numbers are indicated by Roman numerals.

Expression and Subcellular Distribution

Although 34 CDPK genes are predicted in the completed Arabidopsis genome, the transcription of all of these genes has not been demonstrated (Table I). Full-length cDNAs have been obtained for AtCPKs 1 through 12 (Harper et al., 1993; Urao et al., 1994; Hong et al., 1996; Hrabak et al., 1996); for AtCPKs 15, 16, and 21 (Harmon et al., 2001); and AtCPKs 13, 17, and 23 (http://signal.salk.edu). Twelve of the remaining 16 CDPKs either have corresponding EST clones or have been cloned by the Sheen laboratory (Table I). Whether the AtCPK 14, 18, 19, and 31 genes are transcribed is still unknown. Therefore, there are at least 30 distinct CDPKs expressed in Arabidopsis.

The subcellular localization of only three Arabidopsis CDPKs has been studied. All three of these proteins are associated with membrane systems, with AtCPK2 localized to the ER (Table I; Lu et al., 2001, 2002). In other species, CDPKs have been found associated with both soluble and membrane fractions, nuclei, chromatin, and the cytoskeleton (e.g. Putnam-Evans et al., 1989; Schaller et al., 1992; Martin and Busconi, 2000; Patharkar and Cushman, 2000). The plasma membrane localization of some CDPKs is highly significant because many critical calcium-signaling events are initiated by fluxes across the plasma membrane (Malhó et al., 1998; Rudd and Franklin-Tong, 2001).

	REGULATION OF CDPK ACTIVITY

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

Regulation by Phosphorylation and Dephosphorylation

In addition to Ca²⁺, reversible phosphorylation also may regulate CDPK kinase activity. Autophosphorylation has been observed in both native and recombinant CDPKs (Saha and Singh, 1995; Chaudhuri et al., 1999; Harmon et al., 2000). However, the role of autophosphorylation in the activities of CDPKs is unclear. For example, in vitro autophosphorylation activates a groundnut (Arachis hypogea) CDPK but inhibits one in winged bean (Psophocarpus tetragonolobus; Saha and Singh, 1995; Chaudhuri et al., 1999). Furthermore, autophosphorylation shows no effect on the Ca²⁺ dependence of groundnut and soybean CDPK activities (Chaudhuri et al., 1999; Harmon et al., 2000). CDPK activation may be modulated by other protein kinases as well. For example, full activation of a tobacco (Nicotiana tabacum) CDPK (NtCDPK2) requires both calcium and phosphorylation. Because the phosphorylation event is insensitive to the calmodulin and CDPK antagonist W7 in vitro, and in vitro autophosphorylation cannot substitute for the in vivo phosphorylation-dependent activation, autophosphorylation likely is not responsible for this activation (Romeis et al., 2000, 2001). Therefore, NtCDPK2 is probably activated through direct phosphorylation by an upstream protein kinase. How this phosphorylation event affects CDPK activation is still unknown.

Dephosphorylation is as important as phosphorylation in controlling signaling pathways. A soluble phospho-Ser phosphatase from winged bean shoots dephosphorylates an inactivated, autophosphorylated winged bean CDPK1 (WbCDPK1) in vitro (Ganguly and Singh, 1999). It is thought that this action releases an inhibitory effect of autophosphorylation and suggests the existence of a regulatory feedback loop. These findings indicate an intricate interplay between protein kinases and phosphatases in regulating some CDPK activities.

Regulation by Phospholipids

Biochemical analysis has revealed that in the presence of Ca²⁺, specific phospholipids can enhance in vitro substrate phosphorylation by CDPKs from oat (Avena sativa), Arabidopsis (AtCPK1), and carrot (Daucus carota; DcCPK1) by 2 to 30 times above that observed with Ca²⁺ alone (Schaller et al., 1992; Harper et al., 1993; Binder et al., 1994; Farmer and Choi, 1999). Interestingly, the phospholipids regulating kinase activity are not the same for each of the CDPKs studied, which may provide an added layer of CDPK specificity (Farmer and Choi, 1999). At least two mechanisms for phospholipid activation of CDPKs likely exist. Although both phosphatidylinositol and lyso-phosphatidylcholine increase substrate phosphorylation by AtCPK1, only the prior enhances CDPK autophosphorylation and partially relieves inhibition by poly-Lys (Binder et al., 1994). Sequence analysis has revealed a putative phosphatidylinositol binding site in the N terminus of AtCPK1 (Binder et al., 1994). Because this phospholipid still activates a truncated AtCPK1 protein lacking this site, however, this cannot be the only phosphatidylinositol-binding site in the protein.

It remains unknown whether phospholipids will be important in regulating CDPK activities in vivo. Because of the specificity of the phospholipids stimulating each CDPK, the effects will likely have physiological relevance. Some of these phospholipids do act as second messengers in plant signal transduction, and, therefore, might elicit their effects in part through CDPKs. In addition, the activation of DcCPK1 by phosphatidyl-Ser, which is known to activate protein kinase C (PKC) in animal cells, suggests CDPKs may be responsible for some PKC-like responses in plant cells (Farmer and Choi, 1999). Whether there is a plant PKC homolog is still unknown. This observation also may help explain reports of kinases seemingly having properties of both CDPKs and PKC (Abo-el-Saad and Wu, 1995; Van der Hoeven et al., 1996; Lino et al., 1998).

Regulation by 14-3-3 Proteins

Three different 14-3-3 isoforms have been demonstrated to specifically bind and activate AtCPK1 in vitro in the presence of Ca²⁺ (Camoni et al., 1998b). Calcium may be needed in part to induce autophosphorylation of the CDPK, because 14-3-3 proteins typically regulate the activities of many enzymes by binding specific phosphorylated residues. Although the specific sites of autophosphorylation and 14-3-3 protein interaction for AtCPK1 are unknown, AtCPK1 does contain one site within the N terminus similar to the most common 14-3-3 consensus binding site, R-S/T-X-S-X-P, where the underlined Ser is phosphorylated (Camoni et al., 1998b). Using The Arabidopsis Information Resource Patmatch to search the AtCPKs for similar sites revealed that only AtCPK24 (amino acids 28-33) and AtCPK28 (amino acids 40-45) also possess such putative 14-3-3 binding sites. The putative binding sites for all three proteins are located in the N-terminal variable domain. Whether 14-3-3 proteins bind these sites remains to be determined, but these results nonetheless suggest that 14-3-3 proteins may be regulators of only a subset of CDPKs in Arabidopsis.

	PHYSIOLOGICAL FUNCTIONS

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

Although CDPKs have been implicated biochemically to act as key regulators of many signaling pathways, very little is known about which particular CDPK acts as the calcium sensor in each case. The absence of specific inhibitors for CDPKs (there are general inhibitors), the lack of dominant negative constructs, and the possibility of functional redundancy have made it difficult to assign functions to individual CDPKs. Nevertheless, significant progress has been made in our understanding of the physiological roles of CDPKs in Arabidopsis and other plant species.

Hormones

Changes in the cytosolic Ca²⁺ concentration are known to occur during hormone signaling (Bethke et al., 1995; Bush, 1995), and CDPKs may act as the calcium sensors modulating these responses. For example, hormone-induced increases in CDPK activity have been demonstrated in gibberellin-treated rice seeds (Abo-el-Saad and Wu, 1995), and in brassinolide-treated rice green lamina (Yang and Komatsu, 2001). Significantly, CDPKs have also been shown to activate hormone-responsive genes in vivo. Expression of a constitutively active AtCPK10 or AtCPK30 directly activates a barley (Hordeum vulgare) promoter induced by abscisic acid (ABA) in the absence of stress signals (Sheen, 1996). CDPKs also are regulated at the transcriptional level by hormones. GA, ABA, and cytokinin all induce the expression of NtCDPK1 mRNA in detached tobacco leaves (Yoon et al., 1999), whereas indole-3-acetic acid stimulates the expression of specific CDPK genes in mung bean (Vigna radiata) cuttings (Botella et al., 1996) and in cultured alfalfa (Medicago sativa) cells (Davletova et al., 2001).

Growth and Development

A wide variety of growth and developmental processes in plants are regulated by Ca²⁺ fluxes (Evans et al., 2001; Hepler et al., 2001). One such process involving CDPKs is pollen tube growth. In maize, the expression of a pollen tube-specific CDPK gene is restricted to the late stages of pollen development (Estruch et al., 1994). The addition of a calmodulin antagonist, CDPK inhibitor, or antisense oligo-nucleotides directed against the maize CDPK mRNA impair pollen germination and tube growth. Further, it has been shown that localized increases in cytosolic Ca²⁺ concentration leading to pollen tube reorientation in Agapanthus umbellatus also increase CDPK activities (Moutinho et al., 1998). The ability of a pollen tube CDPK from Nicotiana alata to phosphorylate a style self-incompatibility RNase suggests a role for CDPKs in self-incompatibility regulation, as well (Kunz et al., 1996).

CDPKs may also regulate other developmental processes. In the early stages of potato (Solanum tuberosum) tuberization, the spatial and temporal accumulation of StCDPK mRNA correlates with the increased activity of the protein (MacIntosh et al., 1996; Raíces et al., 2001). In addition, CDPKs may modulate nodulation because two nodule-specific proteins, nodulin-26 (a voltage-sensitive ion channel) and nodulin-100 (a Suc synthase [SuSy]), are phosphorylated by CDPKs in vitro (Weaver and Roberts, 1992; Zhang and Chollet, 1997). Other processes likely involving CDPKs include embryogenesis, seed development and germination in sandalwood (Santalum album; Anil et al., 2000), and sexual organ development in liverwort (Conocephalum conicum; Nishiyama et al., 1999).

Guard Cells and Stomatal Movements

Stimulus-triggered calcium oscillations regulate the guard cell ion fluxes necessary for driving stomatal movements (McAinsh et al., 1997; Assmann and Wang, 2001; Schroeder et al., 2001a, 2001b). During ABA-induced stomatal closure, calcium enhances K⁺ efflux in part via the direct inhibition of inward-rectifying K⁺ channels localized within the plasma membrane (McAinsh et al., 1997; Schroeder et al., 2001a, 2001b). A Vicia faba CDPK from guard cells is capable of phosphorylating the Arabidopsis guard cell voltage-gated K⁺ influx channel KAT1 in vitro in a calcium-dependent manner. This phosphorylation requires the integration of KAT1 into a membrane, suggesting that this regulation is likely to occur in vivo (Li et al., 1998). Co-injecting Xenopus laevis oocytes with mRNAs of KAT1 and a soybean CDPK inhibits K⁺ currents typically seen when injecting KAT1 transcripts alone (Berkowitz et al., 2000). These results suggest that stimulus-induced calcium fluxes in guard cells may activate a CDPK that phosphorylates an inward-rectifying K⁺-channel, down-regulating its activity and resulting in a net efflux of K⁺ from the cell and stomatal closure.

Actually, Ca²⁺ influxes are implicated in hormone and blue light-induced stomatal opening, as well (McAinsh et al., 1997; Schroeder et al., 2001a), and CDPKs may also be important calcium sensors in these cases (Pei et al., 1996). During stomatal opening, anion uptake into vacuoles is necessary to balance K⁺ uptake. A vacuolar chloride channel in V. faba guard cells is highly activated by AtCPK1 in the presence of calcium. AtCPK1 also induces malate uptake into V. faba guard cell vacuoles and Cl uptake into red beet (Beta vulgaris) root vacuoles (Pei et al., 1996). These data suggest that CDPKs may be important in the general regulation of vacuolar anion uptake in plant cells. It is not known if AtCPK1 is directly phosphorylating the ion channels or acts on an intermediate protein.

Carbon and Nitrogen Metabolism

In response to various environmental and growth and developmental changes, plants must modify aspects of carbon and nitrogen metabolism. Two key enzymes of carbon metabolism modulated by CDPKs are SuSy and Suc-phosphate synthase (SPS), which catalyze Suc cleavage and Suc synthesis, respectively (Huber and Huber, 1996). In vitro experiments with SuSy from several plant species have shown that the primary CDPK phosphorylation site is an N-terminally located Ser between residues 11 and 15 (Huber et al., 1996; Nakai et al., 1998; Zhang et al., 1999; Loog et al., 2000). Phosphorylation of this site appears to be CDPK specific because the residue is not phosphorylated by a heterologous SuSy protein kinase (mammalian protein kinase A) (Zhang et al., 1999; Loog et al., 2000). In many species, including Arabidopsis, this Ser and the surrounding amino acids required for CDPK phosphorylation are highly conserved, suggesting such phosphorylation of SuSy may be important for the regulation of Suc metabolism (Huber et al., 1996; Zhang et al., 1999). It has been shown that this phosphorylation reduces membrane association, increasing the amount of cytosolic SuSy (Winter et al., 1997; Zhang et al., 1999). The physiological significance of this phosphorylation-driven change in subcellular localization is not clear, but may be important for directing the use of the cleavage products by specific biosynthetic pathways (Zhang et al., 1999).

Dark inactivation of spinach (Spinacia oleracea) leaf SPS is dependent upon phosphorylation of Ser-158 (McMichael et al., 1993, 1995a, 1995b; Huang et al., 2001). Ser-158 is phosphorylated by a 45-kD CDPK (PK_I) and a 150-kD multiple subunit SNF1-related protein kinase (SnRK1 or PK_III) in vitro (McMichael et al., 1995a, 1995b; Huang and Huber, 2001). However, a Pro at position P-4, as is observed in many dicots besides spinach, greatly reduces the ability of CDPK, but not SnRK1, to phosphorylate a synthetic peptide designed from this site. This suggests a more important role for calcium signaling in the phosphorylation-based inactivation of SPS in spinach and many monocots than in most dicots (Huang and Huber, 2001). CDPKs can phosphorylate the homologous site in maize SPS, which lacks a Pro at this site (Huber et al., 1995). The Arabidopsis SPS possesses this Pro, likely preventing phosphorylation by CDPKs in vivo (Huang and Huber, 2001).

Interestingly, it appears that there is an overlap in the specific kinases able to phosphorylate SPS and a rate-limiting enzyme of nitrogen metabolism-nitrate reductase (NR)-in vitro, indicating a possible central role for CDPKs in the coordination of carbon and nitrogen metabolism in planta. Similar to the phosphorylation of Ser-158 in SPS, phosphorylation of Ser-543 in NR takes place in the dark and results in inactivation (Kaiser and Huber, 2001; MacKintosh and Meek, 2001). However, such inhibition of NR activity additionally requires the binding of 14-3-3 proteins to the phosphorylated residue (Douglas et al., 1995; Bachmann et al., 1996). The CDPK PK_I phosphorylates both SPS and NR (McMichael et al., 1995a; Bachmann et al., 1996; Douglas et al., 1997, 1998). NR also appears to be inactivated through phosphorylation of the same site by an additional CDPK (PK_II) that has little effect on SPS (McMichael et al., 1995a; Bachmann et al., 1995, 1996). Importantly, these different protein kinases may allow for the activities of SPS and NR to be coordinately or separately regulated, depending upon the needs of the cell. Thus, carbon and nitrogen metabolism can be similarly or differentially modulated.

Little is known about the molecular identity of the CDPKs regulating metabolism and how they are activated in vivo. Suc is known to induce cytosolic calcium influxes (Furuichi et al., 2001) and to increase the protein expression and autophosphorylation of a plasma membrane-associated CDPK from tobacco leaves (Iwata et al., 1998). Because SuSy is important for Suc cleavage in sink locations, Suc induction of the CDPK acting on SuSy might take place in this case. Dark-induced CDPK inactivation of SPS and NR may be explained in part by a reported increase in cytosolic calcium in the absence of photosynthesis (Miller and Sanders, 1987; Johnson et al., 1995). In all cases, a direct demonstration of CDPK activation and substrate phosphorylation in vivo will be necessary to fully demonstrate physiological relevance. In addition, the relative importance of CDPKs and calcium-independent protein kinases, such as SnRK1 for SPS, in phosphorylating such enzymes could be explored further.

Abiotic Stress

Many stress signals, such as wounding, cold, high salinity, and drought, are known to elicit fluctuations in cytosolic Ca²⁺ levels, as well as changes in protein phosphorylation (Bush, 1995; Trewavas, 1999; Knight and Knight, 2001). Several lines of evidence suggest that CDPKs mediate abiotic stress signaling pathways. Transcriptional activation of many different CDPKs by a variety of abiotic stresses has been demonstrated in tissues from diverse species (Urao et al., 1994; Monroy and Dhindsa, 1995; Botella et al., 1996; Yoon et al., 1999; Patharkar and Cushman, 2000; Saijo et al., 2000; Chico et al., 2002). The enzymatic activities of CDPKs also increase in response to these stresses. For example, cold treatments enhance activity of a membrane-bound rice CDPK (Martin and Busconi, 2001). In addition, CDPKs have been shown to elicit abiotic stress responses. Using a protoplast transient expression system, specific CDPKs (AtCPK10 and 30 but not AtCPK1 or AtCPK11) have been demonstrated to activate a stress and ABA-inducible promoter. This result shows the connection of particular CDPKs to specific signaling pathways in vivo (Sheen, 1996). Furthermore, overexpression of a rice CDPK (OsCDPK7) in vascular bundles confers cold, salt, and drought tolerance in transgenic rice (Saijo et al., 2000, 2001). These results demonstrate the usefulness of engineering CDPKs to enhance abiotic stress tolerance in crops.

Pathogen Defense

Extensive studies in various plant/pathogen systems have demonstrated that a cytosolic calcium influx is a crucial early step for the activation of pathogen-induced signal transduction cascades (Xu and Heath, 1998; Blume et al., 2000; Fellbrich et al., 2000; Grant et al., 2000; Nürnberger and Scheel, 2001; Rudd and Franklin-Tong, 2001). Pathogen response pathways are often activated by the interaction between a pathogen-encoded elicitor (such as the Cladosporium fulvum Avr9 peptide) and a corresponding plant-encoded receptor (such as the tomato Cf-9 resistance protein). Recently, a CDPK activated in vivo after a Cf-9/Avr9 gene-for-gene interaction has been identified in Cf-9 transgenic tobacco, suggesting that CDPKs are important calcium sensors in inducible defense responses (Romeis et al., 2000, 2001). The transition from a nonelicited to an elicited form increases enzyme activity by 10- to 200-fold (Romeis et al., 2000). Silencing of tobacco NtCDPK2 and NtCDPK3 (encoding the CDPKs likely responsible), using a viral-induced gene silencing system, has shown that CDPKs are essential for mediating the Cf-9-/Avr9-induced hypersensitive response in planta (Romeis et al., 2001). Significantly, similar results are also observed in the Cf-4/Avr4 gene-for-gene interaction, indicating a more general role for CDPKs in elicitor signaling events. It will be of particular interest to examine whether the Arabidopsis orthologs of NtCDPK2 and NtCDPK3, AtCPK2 and AtCPK1, also are involved in pathogen defense.

The specific function of CDPKs in early defense responses is not yet understood, but, because calcium fluxes and CDPK activation are upstream events of defense signaling (Blume et al., 2000; Fellbrich et al., 2000; Grant et al., 2000; Romeis et al., 2000), they likely regulate other early processes. One of these steps could be the formation of reactive oxygen species by NADPH oxidase, which is both calcium and phosphorylation dependent (Mehdy, 1994; Grant et al., 2000). Although NADPH oxidase may be directly regulated by calcium (Keller et al., 1998), studies have shown that calmodulin/CDPK antagonists can inhibit the phosphorylation and membrane translocation of putative cytoplasmic complex components and block reactive oxygen species formation (Xing et al., 1997; Romeis et al.; 2000). It has been suggested that CDPKs may mediate this indirect calcium regulation (Mehdy, 1994; Xing et al., 1997; Romeis et al., 2000). Recent data revealing that ectopic expression of AtCPK1 can enhance NADPH oxidase activity in a cell-free system and in tomato protoplasts (Xing et al., 2001) support these studies.

CDPKs activated by pathogen elicitors may also initiate other early ion fluxes. For instance, CDPKs may be responsible for changes in H⁺ fluxes resulting from the calcium- and phosphorylation-dependent regulation of plasma membrane proton ATPases (PM H⁺-ATPases; Schaller and Sussman, 1988; Schaller et al., 1992; Camoni et al., 1998a; Lino et al., 1998; De Nisi et al., 1999; Schaller and Oeckling, 1999). Depending on the elicitor, elicitor treatment may increase the net H⁺ influx through negative regulation of PM H⁺-ATPases or enhance the net H⁺ efflux through positive regulation of these pumps (Blumwald et al., 1998). Several studies have shown that purified CDPKs can phosphorylate PM H⁺-ATPases in vitro (Schaller et al., 1992; Harmon et al., 1996; Camoni et al., 1998a). The most detailed report shows that a CDPK from maize roots phosphorylates a PM H⁺-ATPase in the C terminus (Camoni et al., 1998a), phosphorylation of which is thought to be important for 14-3-3 protein-mediated activation of the protein (Morsomme and Boutry, 2000). Further biochemical analysis will be necessary to identify the precise phosphorylation sites and to determine how these phospho-modifications affect the ATPase activity because CDPK phosphorylation of different sites may yield positive or negative regulation.

	CDPK SUBSTRATES

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

Although CDPKs have been shown to phosphorylate a large number of protein substrates in vitro (Table II), limited information is available regarding substrate specificity and phosphorylation sites in vivo. Although CDPKs are highly homologous, an examination of in vitro substrate phosphorylation by spinach and soybean CDPKs suggests that CDPKs will exhibit substrate specificity differences in vivo (Bachmann et al., 1996; Lee et al., 1998). The use of synthetic peptides has facilitated delineation of the potential phosphorylation motifs recognized by some CDPKs. Most of these studies have been conducted with spinach leaf CDPKs PK_I and PK_II, which both appear to recognize two different general phosphorylation motifs (McMichael et al., 1995b; Bachmann et al., 1996; Huang and Huber, 2001; Huang et al., 2001). The first minimal consensus phosphorylation site is ₅-X₄-Basic₃-X₂-X₁-S, where S is the phosphorylated Ser, X is any residue, and  is a hydrophobic residue (McMichael et al., 1995b; Bachmann et al., 1996; Huang and Huber, 2001). Substituting a Thr for the Ser makes the synthetic peptide a poor substrate for PK_II but not PK_I (Bachmann et al., 1996). Further studies of PK_I reveal that basic residues at P  6 and P + 5 are not required but enhance phosphorylation (Huang and Huber, 2001). Therefore, maximal phosphorylation by PK_I is attained with the motif Basic₆-₅-X₄-Basic₃-X₂-X₁-S-X₊₁-X₊₂-X₊₃-₊₄-Basic₊₅. This motif is similar to the site delineated for cauliflower (Brassica oleracea) floret PK_I and maize CDPK-1, except that the maize protein also requires a basic residue at P+2 (Toroser and Huber, 1998; Loog et al., 2000). This recognition site also is similar to that recognized by mammalian PKC (Loog et al., 2000).

Making peptides with various Ala substitutions, a second substrate phosphorylation motif for PK_I and PK_II has been identified as Basic₉-Basic₈-X₇-Basic₆-₅-X₄-X₃-X₂-X₁-S-X₊₁-Basic₊₂ (Huang et al., 2001). Using this motif, several new potential CDPK substrates have been identified, including the spinach plasma membrane aquaporin PM28A, ACC synthase, a splicing factor, the RNA polymerase -chain, a protein kinase, and two disease resistance protein homologs (Huang et al., 2001). Both PK_I and PK_II phosphorylate a synthetic peptide based on the phosphorylation site of PM28A (Huang et al., 2001). Furthermore, PM28A is regulated by calcium-dependent phosphorylation of this site (Johansson et al., 1998), suggesting the physiological relevance of the prediction. These results support the usefulness of predicting CDPK substrates using these motifs. Interestingly, the second motif is not recognized by the spinach SNF1-related protein kinase SnRK1 (PK_III), which has been shown to phosphorylate some of the same substrates as PK_I and PK_II (McMichael et al., 1995a; Douglas et al., 1997; Huang and Huber, 2001). Therefore, using the first phosphorylation motif may allow for phosphoregulation by both CDPKs and SNF1-related protein kinases in vivo, whereas the second motif may allow for sole regulation by CDPKs (Huang et al., 2001).

	CDPK SPECIFICITY

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

To carry out the myriad of Ca²⁺-dependent cellular processes, plants have acquired a multitude of CDPK genes during evolution. How different CDPKs translate the information encoded in the "calcium signatures" to specifically affect metabolism and gene expression is mostly unknown. Available evidence indicates that functional specialization of individual CDPKs can occur through different types of regulation. For example, detailed analysis of three soybean CDPK isoforms has shown conclusively that each one possesses unique Ca²⁺-binding properties, presumably allowing different thresholds of Ca²⁺ activation (Lee et al., 1998). In addition, modulations by reversible phosphorylation (winged bean WbCDPK1; Ganguly and Singh, 1999), different phospholipids (carrot DcCPK1; Farmer and Choi, 1999), and 14-3-3 proteins (AtCPK1; Camoni et al., 1998b) further contribute to CDPK specificities. Targeted subcellular localization, such as to the plasma membrane (rice OsCPK2; Martin and Busconi, 2000), ER (AtCPK2; Lu and Hrabak, 2002), and nuclei (ice plant McCDPK1; Patharkar and Cushman, 2000), allows interactions with different substrates (Table II). Variations in phosphorylation motif recognition or substrate phosphorylation may occur as well (Bachmann et al., 1995, 1996; McMichael et al., 1995a; Lee et al., 1998). Specific roles of CDPKs also are determined by stimulus-specific (e.g. wound-induced tomato LeCDPK1; Chico et al., 2002) and spatiotemporal regulation of expression (maize pollen-specific CDPK; Estruch et al., 1994). Thus, plants may use a combination of various strategies to functionally specialize individual CDPKs, as evidenced by two sandalwood CDPKs isoforms that differ in tissue-specific distribution, subcellular localization, and enzyme kinetics and properties (Anil et al., 2001).

	PERSPECTIVES

TOP ABSTRACT INTRODUCTION THE ARABIDOPSIS CDPK GENE... REGULATION OF CDPK ACTIVITY PHYSIOLOGICAL FUNCTIONS CDPK SUBSTRATES CDPK SPECIFICITY PERSPECTIVES LITERATURE CITED

As we enter the post-sequencing era of Arabidopsis, understanding the physiological roles of the array of Arabidopsis CDPKs poses a new challenge. A combination of biochemical, molecular, cellular, and genetic approaches will be required to elucidate the function(s) of each individual CDPK. The spatial and temporal expression patterns of CDPKs can be determined by using promoter/reporter (e.g. -glucuronidase or green fluorescence protein; Jefferson et al., 1987; Chiu et al., 1996) fusions in transgenic plants and in situ hybridization with isoform-specific probes. The expression of epitope-tagged CDPKs in a transient expression system or transgenic plants can provide information regarding subcellular localization (Chiu et al., 1996; Hwang and Sheen, 2001). New CDPK-interacting proteins and substrates can be identified using bacterial expression, phage display, or yeast two-hybrid techniques (Patharkar and Cushman, 2000; Shinohara et al., 2000). Substrate specificity can be determined by performing protein kinase assays with epitope-tagged CDPKs expressed in protoplasts (Cheng et al., 2001). This approach could be further explored on a genomic scale, using immobilized substrates (proteins or peptides) on microarray surfaces (Zhu et al., 2000). The previously delineated phosphorylation motifs and potential substrates (Table II) could be used as a starting point. Finally, gene disruption and silencing techniques, such as insertional mutagenesis, RNA interference, and virus-induced gene silencing, can be used to study altered phenotypes (Krysan et al., 1996; Waterhouse et al., 1998; Romeis et al., 2001). Sequencing of publicly and privately generated insertion mutants of Arabidopsis has already identified a number of putative CDPK mutants, as listed in Table III, with others rapidly becoming available. Due to possible redundancy in CDPK functions (Sheen, 1996), the simultaneous inactivation of highly homologous CDPKs with similar expression patterns, cellular localization, and substrate specificity may be necessary to reveal a mutant phenotype and to identify physiological functions. This integrated approach, in combination with bioinformatics, should shed light on the role of CDPKs in the complex web of signaling networks that regulate cellular metabolism, growth and development, and responses to the environment. Because specific CDPKs act as positive regulators to selectively activate a stress pathway (Sheen, 1996) and to enhance drought/salt tolerance (Saijo et al., 2000), increasing our understanding of the specific role of CDPKs in many aspects of plant biology will prove invaluable for many future biotechnology applications.